The major justifications for choosing PLS-SEM over CB-SEM relate to sample size, data distribution, and the use of formative indicators. While CB-SEM requires a large sample size (mostly 100-200 or above), PLS-SEM does not require a large sample size to handle even highly complicated models. I have seen papers using PLS-SEM with sample size of 35, 41, 42, etc. Moreover, unlike CB-SEM, PLS-SEM does not require a normal distribution of data. This is a crucial advantage of the PLS-SEM because most of the time, research data fail to follow multivariate normal distribution, and the problem is more pronounced with small samples. Regarding this, this paper may be useful to you:
Hair, J. F., Sarstedt, M., Hopkins, L., and G. Kuppelwieser, V. (2014). Partial least squares structural equation modeling (PLS-SEM) An emerging tool in business research. European Business Review, 26(2), 106-121.
Hi, bold recommendation: Don't use PLS ever. You should take a look into the following very critical discussions in PLS:
Rönkkö, M., McIntosh, C. N., & Antonakis, J. (2015). On the adoption of partial least squares in psychological research: Caveat emptor. Personality and Individual Differences, 87, 76-84.
Ronkko, M., & Evermann, J. (2013). A critical examination of common beliefs about partial least squares path modeling. Organizational Research Methods, 16(3), 425-448. doi:10.1177/1094428112474693
McIntosh, C. N., Edwards, J. R., & Antonakis, J. (2014). Reflections on partial least squares path modeling. Organizational Research Methods, 17(2), 210-251. doi:10.1177/1094428114529165